An architecture for defeasible-reasoning-based cooperative distributed planning

Abstract

Cooperation plays a fundamental role in distributed planning, in which a team of distributed intelligent agents with diverse preferences, abilities and beliefs must cooperate during the planning process to achieve a set of common goals. This paper presents a MultiAgent Planning and Argumentation (MAPA) architecture based on a multiagent partial order planning paradigm using argumentation for communicating agents. Agents use an argumentation-based defeasible reasoning to support their own beliefs and refute the beliefs of the others according to their knowledge. In MAPA, actions and arguments may be proposed by different agents to enforce some goal, if their conditions are known to apply and arguments are not defeated by other arguments applying. In order to plan for these goals, agents start a stepwise dialogue consisting of exchanges of plan proposals to satisfy this open goal, and they evaluate each plan proposal according to the arguments put forward for or against it. After this, an agreement must be reached in order to select the next plan to be refined. © 2011 Springer-Verlag.